# Dynamic Systems Estimation - Multivariate Time Series Package

### Description

Functions for time series modeling, including multi-variate state-space and ARMA (VAR, ARIMA, ARIMAX) models.

### Details

A *Brief User's Guide* is distributed with dse as a vignette.
The package implements an R/S style object approach to time series
modeling. This means that different
model and data representations can be implemented with fairly simple
extensions to the package.

The package includes methods for simulating, estimating, and converting among different model representations. These are mainly in dse. Package EvalEst has methods for studying estimation techniques and for examining the forecasting properties of models. There are also functions for forecasting and for evaluating the performance of forecasting models, as well as functions for evaluating model estimation techniques.

Package: | dse |

Depends: | R, setRNG, tframe |

License: | free, see LICENSE file for details. |

URL: | http://tsanalysis.r-forge.r-project.org/ |

The main objects are:

- TSdata
time series input and output data structure

- TSmodel
a DSE model structure

- TSestModel
model, data and some estimation information

The main general methods are:

- TSdata
create, extract a DSE data structure

- TSmodel
create, extract a DSE model structure

- simulate
simulate a model to produce artifical data

- toSS
convert to a state-space model

- toARMA
convert to an ARMA model

- ARMA
construct an ARMA model

- SS
construct a state-space model

- l
evaluate a model with data

- smoother
calculate the smoothed state estimate

The main estimation methods are:

- estVARXls
estimate an ARMA model with least squares

- estVARXar
estimate an ARMA model with ar

- estSSfromVARX
calculate a state-space model from an estimated VAR model

- bft
a (usually) good “black-box” estimated model

- estMaxLik
estimate a model using maximum likelihood

The main diagnositic methods are:

- checkResiduals
autocorrelation diagnostics

- informationTests
calculate several information tests for a model

- McMillanDegree
calculate the McMillanDegree of a model

- stability
calculate the stability of a model

- roots
calculate the roots of a model

The methods for producing and evaluating forecasts are:

- l
evaluate a model with data (and simple forecasts)

- forecast
calculate forecasts

- featherForecasts
calculate forecasts starting at different periods

- horizonForecasts
calculate forecasts at different horizons

- forecastCov
calculate the covariance of forecasts

- MonteCarloSimulations
multiple simulations

The methods for evaluating estimation methods are:

- EstEval
evaluate estimation methods

The functions described in the
*Brief User's Guide* and examples in the help pages should work
fairly reliably (since they are tested regularly), however, the
code is distributed on an “as-is” basis.
This is a compromise which allows me to make the software
available with minimum effort. This software is not a commercial
product. It is the by-product of ongoing research.
Error reports, constructive suggestions, and comments are welcomed.

### Usage

library("dse")

library("EvalEst")

### References

Anderson, B. D. O. and Moore, J. B. (1979) *Optimal Filtering*.
Prentice-Hall.

Gilbert, P. D. (1993) State space and ARMA models: An overview of the equivalence. Working paper 93-4, Bank of Canada. Available at http://www.bankofcanada.ca/1993/03/publications/research/working-paper-199/

Gilbert, P. D. (1995) Combining VAR Estimation and State Space
Model Reduction for Simple Good Predictions. *J. of Forecasting:
Special Issue on VAR Modelling.* **14**:229–250.

Gilbert, P.D. (2000) A note on the computation of time series model roots.
*Applied Economics Letters*, **7**, 423–424

Jazwinski, A. H. (1970) *Stochastic Processes and Filtering Theory*.
Academic Press.

### See Also

`TSdata`

,
`TSmodel`

,
`TSestModel.object`